The Impact of Cash Budgets on Poverty Reduction in Zambia:
A Case Study of the Conflict between Well-Intentioned
Macroeconomic Policy and Service Delivery to the Poor
Hinh T. Dinh, Abebe Adugna, and Bernard Myers1
Abstract
In the early 1990s, facing runaway inflation and budget discipline problems, the
Zambian government introduced the so-called “cash budget,” in which government
domestic spending is limited to domestic revenue, leaving no room for excess spending.
This paper reviews Zambia’s experience during the past decade, focusing on the impact
of the cash budget on poverty reduction. The conclusion is that after some initial
success in reducing hyperinflation, the cash budget has largely failed to keep inflation
at low levels, created a false sense of fiscal security, and distracted policymakers from
addressing the fundamental issue of fiscal discipline. More important, it has had a
deeply pernicious effect on the quality of service delivery to the poor. Features
inherent to the cash budgeting system facilitated a substantial redirection of resources
away from the intended targets, such as agencies/ministries that provide social and
economic services. The cash budget also eliminated the predictability of cash releases,
making effective planning by line ministries difficult. Going forward, Zambia must
adopt measures that over time will restore the commitment to budget discipline and
shelter budget execution decisions from the pressures of purely short-term exigencies.
World Bank Policy Research Working Paper 2914, October 2002
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the
exchange of ideas about development issues. An objective of the series is to get the findings out quickly,
even if the presentations are less than fully polished. The papers carry the names of the authors and should
be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely
those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors,
or the countries they represent. Policy Research Working Papers are available online at
http://econ.worldbank.org.
1
This paper draws on a longer, more comprehensive report by Dinh et al. (2000). We would like to thank
Yaw Ansu, Heinz Bachmann, Philippe Le Houerou, Charles Humphreys, Allister Moon and Mushiba
Nyamazana for valuable input and comments. Financial support from a Japanese grant (PHRD TF026386)
is gratefully acknowledged.
1
The cash budget in Zambia is a clear example of a well- intentioned
macroeconomic policy that has had an adverse impact on poverty outcomes over the
years. The cash budget was supposed to bring down inflation, restore fiscal discipline,
and contribute to economic growth. Instead, after some initial success in reducing
hyperinflation, it has contributed to keeping inflation at high levels, misallocated
resources, and exacerbated the inefficiencies and inequity in the delivery of social
services. More important, the cash budget has created a false sense of fiscal security and
distracted policymakers from addressing the fundamental budget problem, the lack of
fiscal discipline. Zambia’s long dependency on the cash budget means that any move
away from it needs to be gradual and accompanied by measures to address budget
discipline.
Section I reviews the rationale for the cash budget, describes the law and the
regulations governing the Zambian budget, and describes the working of the cash budget
system in practice in Zambia. The macroeconomic effects of the cash budget are
examined in Section II. Section III details the adverse microeconomic effects of the cash
budget, focusing on the reallocation of resources and reduced efficiency of government
operations. The effect of cash budgeting on over commitments and arrears is explored in
Section IV. Section V highlights the impact of cash budgeting as distinct from poor
budget management generally. Section VI looks at cross-country comparisons of the
impact of cash budgeting, briefly examining the experience in Tanzania and Uganda.
Section VII looks at ways in which the deleterious effects of cash budgeting can be
reduced. Conclusions are drawn in Section VIII.
I.
THE CASH BUDGET SYSTEM IN ZAMBIA
The cash budget was introduced in Zambia in the early 1990s, against a backdrop
of runaway inflation and macroeconomic instability. Repeated attempts at stabilization in
the late 1980s having failed, the newly formed Movement for Multiparty Democracy,
which came to power in 1991, faced a huge external debt overhang, depleted foreign
exchange reserves, and high and accelerating inflation. GDP declined by 2% in 1992, the
debt to GDP ratio reached 210 percent of GDP, and foreign exchange reserves fell to less
than one week’s worth of imports. Inflation averaged 189 percent between 1991 and
1993, and the overall budget deficit (before grants) averaged more than 10 percent of
GDP (Annex Table 1).
On the advice of foreign advisors, the cash budget was introduced in late 1993 as
a way of reducing the budget deficit and restoring fiscal discipline to a new liberal
regime. The rationale was simple: if spending could be kept within the hard constraint of
earned revenue, the deficit would be zero and inflation would disappear.
Legal Basis for the Cash Budget
The legal basis for the cash budget is provided by the Financial Regulations and
Finance (Control and Management Act, Cap. 600) of the laws of Zambia. The so-called
Yellow Book (or approved budget) empowers the Permanent Secretary of the Ministry of
2
Finance and Natio nal Planning to “impose a restriction on expenditure under any subhead or item appearing in the estimates.”2 Under the cash budget system, this authority
was broadly interpreted to mean that the executive arm of government (the Ministry of
Finance and National Planning) could decide which activities were to be funded among
those initially approved by Parliament. As a result, under the cash budget system,
Parliament’s approved budget estimates, assented to by the President through the
Appropriations Act, no longer formed the basis for actual funding of government
operations, providing neither the floor nor the ceiling for government expenditures.
Implementation of the Cash Budget
Implementation of the cash budget begins with the Permanent Secretary of the
Ministry of Finance and National Planning issuing a letter of instruction to the central
bank (the Bank of Zambia) to the effect that funding requests from the Treasury should
not be honored and debit transactions not be posted to the Treasury account unless the
Treasury has sufficient cash in the bank to cover the expenses. “Cash in the bank” is
defined as the net credit balance in the Treasury’s “composite position,” which includes
the consolidated balance in specified accounts. The Treasury is enjoined from borrowing
from the Bank of Zambia or running a net overdraft for the year at the Bank, except in a
few special circumstances (allowing payment of personal emoluments mid- month, to be
repaid before the end of the month, for example). The rules, however, still permit a
deficit at any point during the budget year if cash is raised in advance by borrowing from
outside the Bank of Zambia (by selling Treasury bills, for example, which would not
entail an increase in money circulation and would have no inflatio nary effect).
Despite extreme budget exigencies caused by the adoption of the cash budget, few
major breaches were observed in the beginning, largely because of the strong support
from the Cabinet, Parliament, and the public at large (Bolnick 1995). The policy was
endorsed by the Cabinet, trumpeted publicly in the 1993 Budget Address, and approved
by Parliament as part of the budget program. The new government wanted to demonstrate
credibility, which was essential for reviving the economy and sustaining donor support.
Announcing the cash budget as a foundation of its stabilization policy created a situation
in which the government could not violate the rules without incurring a heavy cost, both
economically and politically. The cash budget also provided the Budget Office with a
welcome tool for turning down the constant barrage of spending requests. Finally, there
was widespread public consensus that the government needed to stop printing money.
This consensus— based on the bitter lessons learned from the previous decade of
stabilization failures—was reinforced by the fact that in 1993 the economy was on the
verge of hyperinflation. It could be argued that the heart of the reform was not so much
the changes in the rules but the political commitment to implementing the new rules.
2
According to Bolnick (1995), the cash budget was introduced administratively without enactment of new
statutes or amendments to the existing ones. He argues that this was appropriate, as Zambia had never had a
good record of implementing Acts of Parliament.
3
Operation of the Cash Budget in Practice
To implement the cash budget system, a Joint Data Monitoring Committee was
created, comprising officials and advisors from the Budget Office and the External
Resource Mobilization Department of the Ministry of Finance and National Planning and
from the Bank of Zambia. The committee is responsible for daily monitoring of cash
budget execution, proper management of reserve money, and overall adherence of
macroeconomic policy to program targets. The committee meets daily to scrutinize fiscal
and monetary data, analyze near-term prospects, identify substantive problems, discuss
alternative solutions, and recommend appropriate remedies. It submits weekly reports to
the Minister of Finance, the Governor of the Bank of Zambia, an interministerial
technical group on economic policy, and other top officials. In this way, critical issues are
promptly reported to the top economic policymakers.
The aggregate amount that can be released each month is the difference between
the revenue projection of the Zambia Revenue Authority (called the “revenue profile”)
and the monthly surplus necessary to meet the IMF’s quarterly Enhanced Structural
Adjustment Facility (ESAF) benchmark. The decision on how much each ministry and
other budget head can spend each month in each expenditure category is made ad hoc by
a small committee within the Ministry of Finance and National Planning, comprising the
Permanent Secretary (Budget and Economic Affairs), the Director of the Budget, the
Chief and Principal Budget Analysts, the Chief Revenue Analyst, and the AccountantGeneral (or a representative).
The committee applies a general set of criteria in its decision making. First
priority is given to domestic debt service, contingency-set-aside for unexpected revenue
shortfalls, and personal emoluments, based on the payroll for the previous month.
Servicing of foreign debt is financed overwhelmingly by foreign non-project aid and as
such remains largely outside the cash budget process, which deals with the domestic
budget. 3 Second priority is given to social sector expenditure. Funding of the remaining
expenditures—recurrent departmental charges (known as “operations and maintenance”
in other countries), grants and other current expenditures, and capital expenditure—is
decided last, as a residual, with the small amount of domestically financed capital
expenditures usually receiving the lowest priority. In consequence, recurrent
departmental charges and capital expenditures for low-priority budget heads often receive
no funding during a given month if revenue is particularly low or if new, higher-priority
funding requests are introduced that were not included in the Yellow Book.
As the committee progresses in determining cash releases for the month, cash is
gradually released and ministries and other budget heads are informed of their monthly
resource envelopes. The actual timing of releases varies by type of expenditure. Domestic
debt service is strictly paid at the due dates, and cash releases are structured accordingly.
Personnel emoluments are generally paid by the middle of the month. However, if at that
3
In 2002, the firewall between the domestic budget and the budget for servicing foreign debt was
eliminated.
4
time there is insufficient cash available in the government’s account with the Bank of
Zambia, the Budget Office can draw on its interest- free overdraft there, repaying the
overdraft at the end of the month.
Cash releases for other expenditure categories are usually determined and
executed late in the month, when revenues are at their peak and the risk of an unexpected
revenue shortfall is at a minimum. This means that, de facto, cash releases for purposes
other than debt service and personnel emoluments are not based on the monthly revenue
profile but depend largely on revenue already collected and “in the bank” before the
release takes place. It also means that most of the funds released in a given month are
spent the following month. The May cash release for recurrent departmental charges to
the Ministry of Education, for example, is received by the ministry during the last few
days of May and used largely to cover expenditures during June.
II.
MACROECONOMIC EFFECTS OF THE CASH BUDGET
Initially, the cash budget seemed to have a positive macroeconomic impact. A
year after it was introduced, annual inflation fell from 187 to 53 percent (figure 1 and
Annex Table 1). The domestic balance dropped from a deficit of about 4 percent of GDP
in 1992 to a small surplus in 1995/1996.
2
200
180
0
-2
160
1992 1993 1994 1995 1996 1997 1998 1999 2000
140
120
-4
100
-6
80
Inflation (%)
GDP growth (%) and fiscal
deficit measures (% of
GDP)
Figure 1. Selected Macroeconomic Indicators for Zambia, 1992-2000
60
-8
40
-10
20
Domestic Budget Balance, cash, % of GDP
Overall Budget Surplus/(Deficit), cash, % of GDP
Inflation (end-of-period), %
Several factors cast doubts on whether the cash budget was truly responsible for
these improvements, however. First, at the time the cash budget was introduced, Zambia
also adopted a strict stabilization program in order to restore good standing with the IMF.
Second, following the initial decline, both the budget deficit and inflation ha ve remained
more or less constant. In fact, evidence shows that the overall budget deficit increased
because of quasi- fiscal activities and that these activities were hidden by the false sense
5
of security provided by the cash budget. Third, and most important, the cash budget was
supposed to have worked on the domestic budget, not on the entire budget, which affects
inflation. This turns out to be an important distinction, as explained below.
Until 2002, the cash budget applied to the domestic budget only. If the external
budget was not self- financing as assumed because donor balance of payments support fell
short of the high external debt service payment, maintaining the cash budget was not
sufficient to stabilize the growth of reserve money arising from the claims on the
government (because of bridge loans from the central bank). Because of the lack of fiscal
flexibility—the ability to temporarily change the fiscal balance, for example—the
targeting of the supply of base money through the cash budget made it difficult to offset,
even moderately, the balance of payments impact of external shocks and to target or
maintain a nominal exchange rate simultaneously. As the recent Public Expenditure
Review noted, during adverse external shocks—such as those that occurred in 1995,
1997–98, and 2000 as a result of copper production problems, declines in the price of
copper, aid shortfall, oil price shocks, and other factors—the private sector anticipated
devaluation during the run-down of reserves and moved out of domestic assets into
foreign assets, exacerbating the effects of the original shocks.
Since 2002, the cash budget has continued to operate, but in a broader scheme
where external resources (foreign grants and credits) and debt service payments are
included. Preliminary results indicate that the fluctuations in cash releases have
worsened as external resources have proved to be much more volatile than domestic
revenues.
At this point, we need to clarify the impact of cash budgeting from poor budget
management. It could be argued that by and of itself, the cash budgeting system is not
responsible for the distortions created. After all, in a crisis situation, cuts are often made,
even in countries without cash budgeting systems. So to what degree is the problem
worse in Zambia and to what degree it is caused by cash budgeting as opposed to
unrealistic budgeting? The point to make here is that the problem in Zambia exists every
year, often when the actual overall level of expenditures exceeded the planned amounts,
indicating that the problem is not one of shortage of resources relative to what was
planned, but rather a reshuffling of priorities within the year, facilitated by the cash
budgeting procedures. Of course, this may happen in the absence of a cash budget. But a
discretion based system is certainly more liable to ad-hoc requests, especially from the
highest authorities. Overspending beyond the voted budget is no more part of a cash
budget than of any budget system. But it destroys the raison d’etre of the cash budget in
the first place.
III.
MICROECONOMIC EFFECTS OF THE CASH BUDGET
While the program of fiscal adjustment may have had a positive effect on overall
fiscal and monetary stability, it had a clearly negative effect on the quality of fiscal
management, as became increasingly apparent over time. Budget preparation continued
to suffer from unrealistic budgeting, poorly trained staff, and lack of accountability.
Facilitated by the mechanics of the cash budget process, these problems led to a marked
6
deterioration in budget discipline. Monthly (even weekly) cash releases determined ad
hoc by a small committee in the Ministry of Finance and National Planning very rapidly
replaced the formally approved annual budget as the key instrument for determining and
controlling the allocation and use of budgetary resources. In the process, the system of
laws and rules was replaced by improvisation and muddling through, and transparent
spending decisions endorsed by Parliament based on well thought-out long-term priorities
were replaced by ad hoc decisions made without publicity, dictated by the needs of the
day.
The current cash budget system affects the quality of government operations in three
major ways:
•
It hampers the efficient use of budgetary resources by creating large fluctuations and
great unpredictability in the monthly cash releases to ministries, making it virtually
impossible for managers to plan their activities more than a few weeks ahead or to
undertake major tasks, programs, or campaigns at the most appropriate time of the
year.
•
It results in a suboptimal allocation of resources by encouraging and facilitating the
systematic shift in resources from economically and socially relevant ministries to
general public administration and from recurrent departmental charges to wages.
•
It leads to higher prices charged by government suppliers as a result of growing
overcommitments, arrears, and payment delays.
Before we discuss these consequences, we will present evidence on and causes of
unpredictability in cash releases.
Evidence of Unpredictable Fluctuations in Cash Releases
The extent and frequency of the fluctuations in cash releases is glaring, as
examination of the monthly cash releases for selected expenditure categories received by
four ministries (Health, Education, Agriculture, and Works and Supply) in 1998 reveals
(Table 1). At the time, two of these ministries, Health and Agriculture, had Sector
Investment Programs that required the properly timed release of counterpart funding. The
Ministry of Health, for example, required stable and predictable monthly recurrent
expenditure funding to plan and implement its primary, clinical, and public health
programs.
Table 1. Cash Releases to Selected Ministries for Key Expenditure Categories, 1998
(billions of kwacha)
Ministry of Health
Month
January
February
March
Personal
emoluments
3.1
2.4
3.8
RDCs
1.3
1.2
2.2
Grants
4.4
4.9
3.9
Ministry of Education
Personal
emoluments
7.3
7.3
7.3
7
RDCs
2.1
2.0
1.6
Grants
2.0
2.0
2.0
Ministry of
Agriculture
Ministry of
Works and
Supply
RDCs
1.6
1.2
0.6
Capital
expenditures
0
3.1
0.2
April
May
June
July
August
September
October
November
December
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
1.1
0
1.2
0.9
0.4
4.6
0.7
2.4
5.2
4.2
2.0
9.3
4.0
3.1
5.0
5.1
7.1
7.3
7.3
7.3
7.5
7.8
8.8
10.8
7.8
7.8
4.1
2.4
0.3
2.2
1.5
2.6
3.0
4.8
1.9
2.1
1.5
0.7
1.8
2.1
1.1
1.8
2.1
3.8
0.8
2.1
0
0
0.8
0.5
2.8
2.0
0.3
5.2
0.2
3.7
5.8
5.1
6.1
9.9
4.4
7.2
Budget a
2.0
1.5
4.8
7.9
2.4
1.9
1.0
4.2
Std Dev
0.58
1.21
1.87
1.02
1.18
0.74
0.89
3.02
Note: RDCs: Recurrent Departmental Charges. a. One-twelfth of total original budget appropriation. Standard
deviations are relative to the average budget.
Source: Republic of Zambia Financial Report for the Year Ended 31st December 1998.
Personal emoluments and recurrent departmental charges. Monthly cash
releases for personal emoluments fluctuated relatively little for the Ministries of Health
and Education, with the exception of a sudden decline in April of nearly 50 percent for
the Ministry of Health and an unexplained 30 percent jump in October for the Ministry of
Education (table 1). Cash releases for recurrent departmental charges, however,
fluctuated widely and erratically for the Ministries of Health, Education, and Agriculture.
In the most extreme case, the Ministry of Agriculture received virtually no cash releases
during June and July. These shortfalls were at least partially compensated for by two to
three months of very large cash releases, exceeding 150 percent of original budget
appropriations.
Recurrent departmental charges being the “bread and butter” of government
operations, it is easy to imagine the havoc these large, erratic, and unpredictable
fluctuations create in line ministries and other government agencies. Without a steady
and reliable supply of goods and services, it is difficult to run an administration
efficiently. The timely and predictable availability of many different inputs is vital for
any public administration to produce services reliably and cost-efficiently; sudden
interruptions in their supply will inevitably damage service delivery, often with grave
consequences. In Zambia, for example, rural health administrators had to interrupt
inoculation programs, which they restarted later at much higher costs. Schools had to
suddenly stop meal programs. Agricultural extension service agents had to interrupt farm
visits at critical times of the year for lack of vehicle spare parts, only to resume at much
less opportune times. And the Accountant General’s Office had to postpone longplanned controls of government offices in the hinterland to a much less propitious time
for lack of gasoline.
Monthly recurrent departmental charge releases for the government as a whole
also fluctuated widely, not only in 1998 but also in 1996 and 1997. Such cash releases
fluctuated by more than 50 percent during nearly half of these 36 months. Total monthly
cash releases to the Ministry of Education also fluctuated widely during 1996, 1997, and
1998, varying almost as widely as cash releases for recurrent departmental charge alone.
8
Grants and other payments. For grants and other payments, fluctuations in
monthly cash releases have been less pronounced, albeit still substantial. The coefficient
of variation reached 0.23 for the Ministry of Education and 0.26 for the Ministry of
Health. The affected institutions (universities and hospitals) suffered badly from the
delayed receipt of vital funding and complained loudly about it.
Capital expenditures. For the Ministry of Works and Supply, capital
expenditures in its Roads Department for road maintenance and rehabilitation constitute
by far the most important expenditure item, accounting for well over 90 percent of its
total budget. This kind of work needs to be planned more than a few weeks ahead; it is
disrupted by abrupt cash release fluctuations, which are also costly.
Cash release fluctuations to this department were among the worst, with a
coefficient of variation of 0.55 and monthly cash releases that ranged from zero in
January to K9.9 billion in October. These figures do not refer to expenditure
commitments that could be expected to fluctuate widely when major new projects come
on stream but to authorizations for actual payments, which should be fairly smooth in
tune with the gradual progress in project implementation. These findings are consistent
with the account circulating in the Ministry of Finance and National Planning about the
road project aborted for lack of funds at a critical junction eve n though the total amount
of cash eventually released would have been sufficient to complete it. According to these
accounts, the money was not there when it was needed; when it finally arrived, the
project had already died.
Capital expenditures in most other budget heads also fluctuated widely. Since
these expenditures are financed largely by foreign aid, however, these fluctuations are
less a result of cash budget mechanisms than of the unpredictability of foreign aid.
Causes of Unpredictable Fluctuations in Monthly Cash Releases
The existence of large and unpredictable fluctuations in monthly cash releases is
indisputable. The underlying causes are less obvious. Are the fluctuations the result of
fluctuations in revenues, or are other factors at work?
Fluctuations in revenues. Given the close link between monthly government
revenues and monthly cash releases established by the cash budget system, one would
expect cash release fluctuations to be caused primarily by fluctuations in revenue.
However, Government revenues actually fluctuated less than cash releases, particularly
for recurrent departmental charges.
Revenue fluctuations were also more predictable than fluctuations in cash
releases. These fluctuations followed a clear seasonal trend, remaining low during the
first quarter (in all four years February showed the lowest revenue collection of the year),
rising during the next three quarters, and peaking in December (in three of the four years,
revenue collection was highest in December) (figure 2).
9
Fiscal statistics also indicate that the government has been very successful in projecting
Figure 2. Quarterly Revenue and Expenditure in Zambia, 1995–98
350
300
K'Billion
250
200
150
100
50
0
1995:I 1995:II 1995:III 1995:IV 1996:I 1996:II 1996:III 1996:IV 1997:I 1997:II 1997:III 1997:IV 1998:I 1998:II 1998:III 1998:IV
Year-Quarter
REV
EXP
annual revenues at the time the budget is prepared. During 1995–97, actual revenues
slightly exceeded budget projections; in 1998 the actual shortfall was less than 0.5
percent. Even if the outcome for 1999 is not as striking, this is an impressive track record
(figure 3).
10
During 1995–98 mean monthly revenues were K74.5 billion and mean expenditures were
Figure 3. Monthly Flows of Gross Revenues to the Budget, 1995–98
2.50
2.30
2.10
1.90
In percent of GDP
1.70
1.50
1.30
1.10
0.90
0.70
0.50
0.30
0.10
-0.10
Jan
Feb
Mar
Apr
May
1995
Jun
Jul
1996
Aug
1997
Sep
Oct
Nov
Dec
1998
K66.7 billion. The correlation between monthly revenues and expenditure was about 67.6
percent (figure 4).
Many economic time series based on monthly or quarterly data, such as revenue
and expenditure, exhibit seasonal patterns. A test for seasonality in the revenue and
expenditure data was therefore carried out to remove the seasonal factor and concentrate
Figure 4. Monthly Revenue and Expenditure in Zambia, 1995–98
140
120
80
60
40
20
Se July
pte
mb
No er
ve
mb
er
Ma
y
Se July
pte
mb
No er
ve
mb
er
Ja
n-9
8
M
arc
h
Ma
y
Se July
pte
mb
No er
ve
mb
er
Ja
n-9
7
M
arc
h
Ma
y
Se July
pte
mb
No er
ve
mb
er
Ja
n-9
6
Ma
rch
Ma
y
0
Ja
n-9
5
Ma
rch
K'billions
100
Month
Revenues
Expendit.
on the other components, particularly the trend. To investigate seasonality in the data, we
use one of the simplest methods—the dummy variable approach. If seasonality is present
11
in the data, we would expect the estimated coefficients on the dummy variables to be
statistically significant. The base quarter against which possible seasonality in other
quarters is compared is the first quarter (sometimes known as the omitted dummy).
To investigate seasonality in the monthly data, we run the following regression:
EXPt = α 0 + α 1 REVt + α 3 D2t + α 4 D3t + α 5 D4t + ... + α 13 D12t + u t
where EXPt = expenditure in period t, REVt = revenue in period t , and
Dit = 1 for ith month
= 0 otherwise
The base month for the dummy variables is Janua ry. We do not detect any
seasonality in the monthly revenue data: none of the dummy variables that capture
monthly variations in revenue is statistically significant. Since there is no evidence for
seasonality in Zambia’s monthly domestic revenue, the regression could be rerun without
the dummy variables:
EXPt = 24.10 + 0.57 REV t
(3.39) (6.22)
R 2 = 0.46
Adj. R 2 = 0.44
The following points should be noted. First, an increase in revenue by 1 Kwacha
leads to an increase in expenditure by 0.57 Kwacha. This is lower by 0.20 Kwacha for
every quarterly Kwacha earned. Second, only 44 percent of monthly variations in
expenditures are explained by variations in revenue—much less than the 89 percent for
the quarterly data.
As the regression equation shows, the aggregate budget envelope does not hold
well on a monthly basis (in the sense that the proportion of variation in expenditure
explained by variation in revenue is not close to one). Indeed, less than half the variation
in monthly expenditure is explained by variations in revenue on a monthly basis. In other
words, the implication that variations in expenditure should be fully matched by
variations in revenue does not hold for Zambia on a monthly basis. The cash budget
system is characterized by large fluctuations and a high level of unpredictability in the
monthly release of resources to line ministries than what could be explained by revenue
variations.
12
Clearly, factors other than monthly revenues must have had a major bearing on
monthly cash releases. 4 Three key factors which account for these fluctuations are the
quarterly ESAF target, ad hoc allocations to particular budget heads, and shortshrifting of
“low priority” budget heads . In fact, these other factors were so dominant that in more
than half the months over the period 1995-1998 (55 percent), they more than offset
revenue fluctuations, causing expenditures and revenues to move in opposite directions.
Taking account of the time lag only slightly reduced the frequency of this “perverse”
behavior to 48 percent, still a very high share.
Need to meet quarterly targets. Zambia’s budget must reach the quarterly targets
on net bank claims on the government and on the domestic budget balance of the
government, as agreed with the IMF under various ESAFs. This requirement has caused a
clearly recognizable pattern in monthly expenditures, characterized by relatively high
expenditures during the first two months of each quarter—often exceeding revenues—
followed by sharp cutbacks in the last months, in order to achieve the budget surplus
necessary to meet the IMF targets.
A typical case was the first quarter of 1998: expenditures were high in January
and February, exceeding revenues in both months and resulting in an accumulated deficit
of more than K7 bill (more than 4 percent of revenues) at the end of February. In March
expenditures were cut sharply by more than 16 percent, producing a surplus of nearly
K14 billion for the month, wiping out the deficit, and creating an overall surplus for the
quarter of nearly K7 billion.
When personnel is excluded, the cutbacks for line ministries become even more
apparent. For the government as a whole, recurrent departmental charges were cut by a
third, with charges for the Ministry of Agriculture cut by half. The cash release for road
maintenance and rehabilitation in the Ministry of Works and Supply was cut by nearly 95
percent. The IMF targets were thus met but at a huge cost to some line ministries. The
problem, of course, was not with the targets per se, but rather MFNP’s lack of a
disciplined approach to achieving them.
The events in the first quarter of 1998 were not unusual. Similar scenarios took
place in the second quarter of 1995, the first and last quarter of 1996, and the first and
second quarters of 1997. The “ESAF trough” was particularly pronounced during the
first three quarters of every year and somewhat less in the fourth quarter, when
exceptionally high December revenues (in three of the four years, December revenues
were the highest of the year) made it possible to maintain relatively high expenditures
and achieve IMF targets at the same time. During 1995–98, in two-thirds of the first
three quarters, the last month of the quarter had the lowest level of expenditures
(including payment of arrears), in many cases by far the lowest level. In 8 percent of the
4
Even if for purposes of the above analyses expenditures would include the payment of arrears, the results
would not improve: 49 percent of monthly variations would be categorized as large, and 45 percent of
monthly correlations would be perverse based on a time-lag analysis.
13
months it shared that position with another month; only 25 percent of the time was there
another month with lower expenditures. 5
The disruptive impact of the ESAF trough was tempered somewhat by its
predictability; indeed, it should not have taken long for an experienced public
administrator to realize that it was not a good idea to plan a major program or campaign
in March, June, or September, as the risks of the necessary cash release not coming
forward were very high. But what if for technical reasons March, June, or September
happened to be the best time to undertake a program or campaign? Work may have been
shifted to a less than optimal time, resulting in a loss of efficiency in the form of higher
costs, lower quality and relevance of the planned work, or both. The current cash budget
system indeed raises the danger of work being undertaken not when it should be but
when money becomes available.
Alternatively, work may still have been undertaken at the optimal time (funded by
overcommitments), in the hope that the needed cash release would be forthcoming during
one of the following months. As discussed below, this practice results in widespread
arrears throughout the administration. A cash budget system that smoothed out
systematic fluctuations in cash releases within each quarter would go a long way toward
solving this problem.
Ad hoc allocations to particular budget heads. While the budget discipline
imposed by ESAF on a quarterly basis explains much of the frequent, sharp cuts in cash
releases in March, April, and September, it does not explain the exceptionally high cash
releases often observed during the first two months of each quarter that necessitate these
sharp corrections. If the cash budget rules were strictly applied, these excessive cash
releases would not occur.
Lack of detailed information makes it difficult to fully explain this phenomenon.
Ample evidence suggests, however, that the problem is caused by increases in monthly
cash releases to particular budget heads, submitted to and accepted ad hoc by the cash
release committee, often in complete disregard of original budget estimates.
The root cause of the practice is the virtual collapse of budget discipline and
transparency following introduction of the cash budget system. Adoption of that system
led to the nearly complete demise of the annual budget as the determinant of government
expenditures and guardian of financial discipline and its replacement by monthly cash
releases determined largely on an ad hoc basis with a minimum of transparency.
During the first two months of the quarter, these ad hoc increases are often simply
added to the regular stream of cash releases, explaining the frequent overshooting of
expenditure targets during this period. During the last month of the quarter, with ESAF
targets looming, these additions have to be compensated for by massive cuts elsewhere.
5
This significant intraquarterly seasonality was confirmed by a two tailed t-test run by Stasavage and
Mayo (1999).
14
The result is sharp fluctuations in cash releases to individual budget heads and even
greater fluctuations in particular expenditure categories within individual budget heads.
Cash releases to individual budget heads fluctuate far more than cash releases to the
government as a whole, which rarely vary by more than 30 percent from month to month
and, of course, are never cut to zero.
Monthly cash releases for capital expenditures by the Ministry of Works and
Supply in 1998 fluctuated by more than 30 percent in four-fifths of the months between
1995-98. At the Ministry of Health, cash releases for recurrent departmental charges
fluctuated at least 30 percent in three-quarters of the months; at the Ministries of
Agriculture and Education, these releases fluctuated at least 30 percent in 16 of the 48
months. Together the four ministries experienced four months of zero cash release in one
expenditure category or another.
Under the cash budget system, significant increases in the monthly cash release to
a particular budget head must be compensated for by corresponding decreases in cash
releases to other budget heads, either immediately or at the latest during the third month
of the quarter, when ESAF targets have to be met. The result is large and unpredictable
swings in monthly cash releases to individual budget heads.
Shortshrifting of “low-priority” budget heads. A third important factor
explaining the sharp monthly fluctuations in cash releases is the differences in priority
assigned to different expenditure categories. Statistical analysis of the four most
important discretionary expenditure categories shows a coefficient variation of 0.29 for
personnel emoluments, very close to that for expenditures as a whole (0.26). The
coefficient variation increases to 0.37 for grants, 0.48 for recurrent departmental charges,
and 0.73 for capital expenditures .
Once the fiscal discipline imposed by the budget was abandoned, the floodgates
opened for requests for urgent increases in cash releases. Most of these requests were
motivated by short-term political considerations and immediate needs, rather than vital
long-term needs.
Discretionary expenditures classified as first priority (personal emoluments) are
little affected by the floodgate phenomenon, as they usually have first claim on available
resources, even ahead of the new, ad hoc requests. Non-discretionary expenditures, such
as domestic debt service, are even less affected.
Second-priority budget heads (social services) suffer more, while budget heads
and expenditure categories classified as lowest priority suffer most. These budget heads
are fully exposed to the floodgate phenomenon and usually range well behind the
powerful and politically well-connected budget heads receiving urgent, ad hoc cash
releases over and above their budget estimates. As the “receivers of last resort,” these
budget heads suffer badly even from relatively minor fluctuations in government monthly
revenues. This shortshrifting of third-priority budge heads explains the severe
fluctuations in recurrent departmental charges in the Ministry of Agriculture and in
capital expenditures in the Ministry of Works and Supply.
15
Impact on Service Delivery and Poverty Reduction
Reallocation of Government Resources. The replacement of the voted budget by
the monthly cash rationing system as the main determinant of government expenditures
triggered substantial reallocations of resources during budget implementation. Some
reallocations of budgetary appropriations are common in most countries as the year
progresses and situations change. In Zambia, however, these reallocations are unusually
large and regularly benefit certain spending categories over others. Comparison of
original budget estimates with actual, audited expenditures for 1997 reveals massive and
systematic changes in expenditure priorities during the year (table 2).
In general, these changes are away from ministries and other budget heads that
deal directly with economic and social matters toward general public services (such as
defense, police, and home affairs), away from recurrent departmental charges and grants
toward personal emoluments. This change in focus was reinforced by sharp cuts in capital
expenditures financed by domestic resources. 6 As a result, actual public expenditures
turned out to be substantially less oriented toward development than the budget enacted
at the beginning of the year, ignoring vital long-term actions, such as stimulating
economic diversification, in favor of dealing with more immediate, but much less vital,
issues.
The changes in budget priorities had nothing to do with lack of resources. In fact,
government revenue estimates, as presented in the budget, turned out to be remarkably
accurate. In consequence, total cash releases for the year 1997 came very close to total
original budget estimates, with total actual domestic expenditures (recurrent and capital)
ending up just 8 percent below total original budget appropriations.
The following analysis focuses on two aspects of resource reallocation:
reallocations across groups of ministries and reallocations across expenditure categories
within groups of ministries. Such a cross-analysis is a novelty for Zambia, where regular
fiscal statistics do not lend themselves easily to this kind of assessment. The budget
heads have been divided into three groups: general public services (ministries such as
State House, Defense, Police, and Judiciary); economic services (ministries such as
Energy/Water, Agriculture, and Works and Supply); and social services (ministries such
as Education and Health).
6
Most capital expenditures in Zambia are financed by foreign aid and not channeled through the budget.
They are excluded from this analysis.
16
Table 2. Differences between Budgeted and Actual Expenditures, Excluding Foreign
Project Aid, 1997
General public
services
Economic services
Social services
Item
Original budget (billions of kwacha)
Recurrent expenditures
Wages
Recurrent departmental
charges
Grants
Total current expenditures
Capital expenditures
Movable assets
Projects
Total capital expenditure
Grand total
Total
162.0
79.4
28.1
57.0
79.0
40.7
269.1
177.1
55.6
297.0
18.6
103.7
79.8
199.5
154.0
600.2
6.8
10.6
17.4
314.4
19.1
62.0
81.1
184.8
0.7
24.2
24.9
224.4
26.6
96.8
123.4
1,130.3
29.7
36.7
134.7
38.9
346.5
215.7
9.9
76.3
74.8
248.4
150.1
712.3
17.9
32.1
50.0
126.3
0.3
5.4
5.7
254.1
24.2
47.7
71.9
. 1,037.1
170
96
129
123
94
125
97
119
36
22
23
91
49
55
Actual expenditures ( billions of kwacha)
Recurrent expenditures
Wages
182.1
Recurrent departmental
140.1
charge
Grants
65.4
Total current expenditures
387.6
Capital expenditures
Movable assets
6.0
Projects
10.2
Total capital expenditures
16.2
Grand total
403.8
Deviation (actual expenditures as percentage of original budget)
Recurrent expenditures
Wages
112
106
Recurrent departmental
177
64
charges
Grants
118
53
Total current expenditures
131
74
Capital expenditures
Movable assets
88
94
Projects
96
52
Total capital expenditures
93
62
Note: Grand total figures include “other expenditures”, which in turn include miscellaneous
expenditure, debt service and contingencies.
Source: Republic of Zambia Financial Report for the Year Ended 31st December 1997.
Reallocations across Groups of Ministries. The analysis reveals a strikingly
clear picture: general public services as a whole were allowed to overspend their budget
estimates by 28 percent, while economic services saw their spending cut by nearly one
17
third below budget. In order to comply with agreements with the World Bank, social
service ministries overspent their budgets by 13 percent. As a result, the share of general
public services in total expenditures increased substantially, from 34 percent of budgeted
spending to 50 percent of actual spending. Spending on social services rose from 25 to 31
percent. In contrast, spending on economic services declined from 20 percent to 16
percent 7 .
As development issues, by their very nature, are always long term and rarely of
immediate urgency, they usually lose out in the daily battle for funds once budget
priorities are no longer respected. It is indeed difficult to argue convincingly that
rehabilitation of a rural access road cannot possibly be postponed another month so that
the funds can be used to cover the travel costs of an important diplomatic delegation
abroad. In the kind of ad hoc budget implementation system practiced in Zambia today,
there is a danger that the rural road in question will never be rehabilitated because the
funds are instead used to cover “urgent” needs—items not included in the original budget
estimates.
The result is that immediate funding needs are satisfied, while the country’s longterm development is jeopardized. In a country in urgent need of diversifying its economy
beyond mining, deep cuts in funding for agriculture, tourism, and manufacturing
industries, may seem incomprehensible—and certainly not in the longer-term interests of
development. But such shifts in spending have become a seemingly inevitable outcome
of a process in which ad hoc requests from more powerful ministries tend to take
precedence.
A well-known procedure used by some powerful line ministries to avoid the
negative impact of cash budgeting is to budget for lower-priority expenditures in the
original budget (the Yellow Book), later in the year requesting supplementary estimates
for high-priority spending that the cash release committee would find difficult to turn
down. In this way, regularly allocated cash is used for lower-priority purposes, while
high-priority expenditures are financed through ad hoc special requests that result in
reduced cash releases to some other budget head.
From the point of view of economic and social development, it would be of little
concern, if the shifting of resources were limited to reallocations within each group (that
is, an urgent funding request by a general public service ministry, such as the Ministry of
Home Affairs, would be accommodated by cutting the cash release of another general
service ministry, such as the Ministry of Legal Affairs, for instance). The problem is that
economically and socially important ministries are systematically discriminated against
in favor of general public services.
7
For details, see Dinh et al. (2000).
18
Reallocations across Categories within Groups of Ministries. Reallocation of budge t
priorities affected not only the distribution of budgetary resources among ministries. Of
equal importance was the shift in priorities among different categories of expenditures—
more precisely, the systematic shift in priorities among different expenditure categories
within certain groups of ministries.
Recurrent departmental charges. At first sight, Table 2 suggests that during
implementation of the 1997 budget, personal emoluments and recurrent departmental
charges received nearly equal priority, with both expenditure categories overspending
their budgets by similar magnitudes. A closer look, however, reveals major differences.
Overspending on personal emoluments was a widespread phenomenon—resulting from
the unplanned and unforeseen 40 percent wage increase during the course of the year—
that benefited all three groups of public services. In all three, the majority of important
budget heads (that is, the 14 ministries with expenditures of K9 billion or more)
overspent on wages and salaries. In cont rast, overspending on recurrent departmental
charges was limited largely to general public services: seven out of the nine important
ministries were allowed to overspend, and the group as a whole overspent by a staggering
77 percent. At the same time, the important economic ministries saw their cash releases
for recurrent departmental charges cut by 30–46 percent below budget. Among social
services, the Ministry of Health received a slight increase (5 percent), but spending by the
Ministry of Education was more than 16 percent below budget.
Clearly, original budget estimates were largely and systematically ignored when
determining the monthly cash releases for recurrent departmental charges. Nearly half of
such cash releases differed from the original budget estimate by more than 30 percent (on
average by 132 percent); only about one-fifth were within 10 percent of the original
budget.
One unfortunate result of this reallocation of recurrent departmental charges
among groups of ministries was a significant deterioration in the relation between
wages/salaries and materials/supplies for the economically and socially important
ministries. At the time the budget was put together, it was felt that, to operate efficiently,
economic services should be allocated about K2 of recurrent departmental charge for
every K1 of personal emoluments. As a result of the massive budget cuts for recurrent
departmental charges to these ministries during the course of the budget year, this ratio
declined to just 1.2:1. For socia l services, the ratio declined in about the same
proportion. Except in the unlikely case that these ratios had been badly overestimated
during the budget process, such serious cuts could not but hamper the activities of these
two groups of ministries, which are of key importance to the country’s long-term
economic and social development. These cuts explain the shortage of drugs in rural
health clinics and major hospitals, the lack of funds for school meals, the small number of
Accountant Generals’ Office staff in the provinces, and the decline in the number of
agricultural extension service agents visiting farms. 8 Public services cannot function
8
Actual wage and salary payments by the Ministry of Agriculture were 43 percent above budget, while
actual spending on recurrent departmental charges was 46 percent below budget.
19
properly and efficiently and provide the services they were set up to provide without a
guaranteed minimum supply of goods and services.
From an economic and social point of view, the main concern is not the sharp
increase in spending on recurrent departmental charges by several general public service
departments. The key issue is where the money came from. Total actual recurrent
departmental charges by general public services were K61 billion over budget, not a
trivial sum, accounting for 7.5 percent of total discretionary domestic expenditures in
1997. At the same time, recurrent departmental charges for economical and social
ministries were cut by K20 billion. Domestically financed capital expenditures for
projects in these two groups of ministries were cut by nearly K49 billion. Although there
may not be a direct, ironclad link between these two phenomena, this shift of resources is
difficult to justify from a long-term development perspective.
Grants. A similar phenomenon has affected grants. For the government as a
whole, actual cash releases came very close to the original budget estimate (97 percent).
The total figure was the result of two diametrically opposed trends, however: substantial
overspending by general public services, compensated for by massive (47 percent) cuts in
economic services and minor cuts in social services. Payments to universities and
hospitals, by far the largest grant recipients, were largely in line with original budget
estimates. Major cuts in grants affected drought relief, small industries development, the
tourism board, district councils, and district health boards.
At a time when the economy ought to be diversifying into tourism and small
industries, among other sectors, and focusing more on the hinterland (agriculture,
tourism), many of these cuts are difficult to justify from a long-term development
perspective. While the reshuffling of priorities for grants affected a smaller number of
budget heads 9 than did the changes in recurrent departmental charges, the effect was
nevertheless massive, demonstrating the complete disregard of the original budget as the
main determinant of expenditure priorities.
Domestically financed capital expenditures. Domestically financed capital
expenditures were cut by nearly half in 1997, with cuts of just 9 percent in movable
assets (cars, computers) and cuts of 51 percents in investment projects. The way these
cuts were made shows that the same phenomenon determining recurrent departmental
charges and grants is at work, albeit to a slightly lesser extent. Cash releases to the group
of general public services were left largely intact, while social services were cut to the
bone, falling 77 percent, and economic services were cut 38 percent. As a result, the
relative importance of the three groups changed significantly. In the original budget,
general services as a whole were scheduled to receive slightly more than 14 percent of
total cash releases for capital expenditures; in the event, they received more than 22
percent. In contrast, the
9
For more than 40 percent of the budget heads concerned, actual cash release for grants deviated by less
than 10 percent from original estimates, about the same number as those experiencing deviations of more
than 30 percent.
20
Box 1. Impact of the cash budget on service delivery
Example 1: Payment for Civil Works Contract
Due to the cash budget system, many spending departments in Zambia operate under unpredictable and fluctuating cash
constraints. In many cases, orders are placed for goods and services and payment vouchers are prepared and kept, often
unrecorded, until availability of cash. Once funds become available, a selected number of vouchers are sent for
payment, just enough to exhaust the cash. Other vouchers, ready for payment, remain pending for next release of fund.
The above practice has led to: (i) late payment penalties charged by suppliers and incidence on costs; and, (ii) rent
seeking opportunity due to lack of transparency in the selection of vouchers for payment. Although the small size of
economy and limited market for suppliers can be the reasons for availability of suppliers’ credits to the Government,
evidence indicates that the latter has an impact on the price and/or quality of goods and services, particularly when the
commercial rate of interest is as high as 40%. With respect to the civil works contracts and other competitive based
contracts, the late payment charges are added to the invoices; but these charges are implicit in other invoices A recent
examination of invoices in selected government agencies indicates that the late payment interest and other additions
resulted in almost 100% increase in the payment for the works done. This has been the case in the road sector, for
example. Furthermore, when controls are weak, late payment charges could also be used as means for financing kickbacks by pre-dating invoices (submitting invoice before the work is done). Furthermore, the lack of transparency in the
way vouchers are selected for payment has created a rent seeking opportunity. The adverse implications for costs of
service delivery and government operations on the ground has been enormous.
Example 2: Purchase of Ndeke House by Ministry of Health
In order to reduce costs, the Ministry of Health decided to acquire a building which would accommodate the Ministry
Headquarters and all its departments within Lusaka. In this regard the Ministry identified a building situated along
Cairo Road in Lusaka and on 19th January, 1995, the Zambia National Tender Board authorised the Ministry of Health
to purchase the building at a cost of K2 billion. However, due to delays in effecting the payment, the building was sold
to other interested parties. The Ministry then identified Ndeke House and on 15th November, 1996 entered into an
agreement with Liquidators of the former Zambia Airways Corporation for the purchase of the building. According to
the agreement, the price of Ndeke House was US $3,000,000 of which US $1,500,000 was to be paid on 31st October,
1996 and the balance by 31st December, 1996 failure to which interest at the rate of 10% per annum would be charged
on the outstanding balance. The agreement also provided that the Ministry would be entitled to 50% of the rents of the
building soon after the first payment of US $1,500,000.
A scrutiny of records pertaining to the purchase and occupation of Ndeke House revealed the following: (a)Contrary to
the terms of the agreement which required that the building be paid for by December, 1996, the payments were only
effected between December, 1996 and January, 1998. The funds for purchase of the building came from the District
Basket funding of US $3,061,832; (b) Due to delays in paying the purchase price, amounts of US $55,208 and US
$16,952 in respect of interest on the outstanding balance of purchase price and loss of exchange earnings were paid for
which no loss reports had been raised; (c) No rent had been collected from the tenants of the building and as of May,
1998 a total amount of US $67,500 at the monthly rate of US $2,250 was due for the period from December, 1996 to
May, 1998; (d) Contrary to the purpose for which the building was purchased, as of May, 1998 only the Central Board
of Health was accommodated in the building while the Ministry Headquarters and its Departments were still in rented
accommodation. In this regard amounts totalling K245,748,276 paid as rent between December, 1996 and February,
1998 and K61,684,500 arrears which had not been paid are wasteful expenditure for which a loss report should be
raised.
Source: Informal World Bank Reports by Iraj Talai and Mushiba Nyamazana; and, Republic of Zambia: AuditorGeneral’s 1997 Report.
share of spending allocated to social services plummeted from more than 20 percent in
the budget to a negligible 8 percent. Economic services, which were cut somewhat less
than the average, saw their share increase slightly, from 66 to 70 percent.
As a result of these massive cuts in investment projects, the share of domestically
financed capital expenditures in total domestic government expenditures fell from nearly
11 percent in the original budget to less than 7 percent during implementation. While this
may be regrettable from a long-term development perspective, it is of limited practical
significance, given that more than 75 percent of government investments are financed by
21
foreign donors. As such they are realized outside regular budget procedures and are not
recorded in the budget accounts, even though they are included in the original budget.
Reduced Efficiency of Government Operations. The productive efficiency of
government operations is severely hampered by the large and highly unpredictable
fluctuations in the resources released by the cash budget committee to budget heads every
month. It is the combination of the above factors—replacement of the voted budget by
the monthly cash rationing system, reallocations across groups of ministries,
reallocations across categories within groups of ministries, and above all fluctuation and
unpredictability—that makes the system so damaging. If budget cuts could be
anticipated, plans could be adjusted accordingly and damage could be limited. But cuts
appear to be largely random, making it virtually impossible for ministries to plan their
activities more than a few weeks ahead. Over the years, line ministries and other budget
heads have devised strategies for sur viving in this very hostile environment—some more
legal than others, some more costly than others. All of these strategies further undermine
budget discipline, making this an inefficient and costly way of running a ministry or any
other public administration.
IV.
EFFECT OF CASH BUDGETING ON OVERCOMMITMENTS AND
ARREARS
The primary justification for the cash budget is improvement in fiscal control, but
overcommitment and arrears have continued to be a serious problem in Zambia. Over the
past several years, an increasing amount of funding has gone to pay for these arrears.
Despite this spending, the stock of arrears does not seem to have declined significantly,
as new arrears are created as old arrears are paid off. By detaching cash releases from the
implicit spending ceilings in the Yellow Book, cash budgeting created a context in which
new arrears were virtually inevitable.
In 1995, K22.2 billion, or 3.7 percent of total domestic budget outlays, was spent
on arrears. Since then the amounts have increased annually, reaching K116.8 billion, or
10.3 percent of total domestic budget outlays, in 1998 and K121 billion (8.8 percent of
total domestic budget outlays) in 1999. This means that in 1999, nearly 9 percent of
budget resources was spent to pay for expend itures made in violation of budget
regulations. Other, regular budget expenditures had to be reduced accordingly.
Published fiscal statistics do not indicate who made those overcommitments or for
what purpose. The process is not transparent, and there is no way of knowing if the 9
percent of budget resources spent on paying arrears in 1999 was not an additional means
of changing budget priorities in the undesirable direction discussed above.
The statistics also fail to reveal whether these overcommitments were
overcommitments only with respect to the cash budget (that is, goods were purchased in a
given month over and above the cash release but within the original budget estimate) or if
even the original budget estimate did not provide for the purchase of these goods. In the
second case, the overcommitment has little, if anything, to do with the cash budget,
22
representing instead an extreme case of lack of budget discipline that ought to be
sanctioned rapidly and severely.
Where money is spent in violation of the cash budget release but respecting the
original budget estimate, the situation is less clear legally, morally, and financially. In
certain circumstances, overcommitments could well be the lesser of two evils, with the
main culprit the cash budget system itself. What, after all, is a ministry supposed to do if
it receives no allocation for recurrent departmental charges during two consecutive
months?
If overcommitments do, in fact, represent funds originally included in the budget,
they could be expected to decline significantly once the cash budget process is improved
and the sharp, monthly fluctuations in cash releases for recurrent departmental charges
have become a thing of the past. As long as this defect in the cash budget system persists,
however, some overcommitments will continue to be made. Since overcommitments can
provide an important safety net when insufficient cash for a service is provided in a given
month—making sure that poor children receive their school meals, for example—it may
make sense to tolerate them as long as they respect the limits imposed by the original
budget.
V.
DISENTANGLING THE IMPACT OF CASH BUDGETING FROM POOR
BUDGET MANAGEMENT
A valid question to ask is whether the deviations observed between budget and
actual expenditure are really due to the cash budgeting process and not due to more
generic problems in budget management. Three important points should be noted in this
regard. First, one might argue that a cross-country analysis can shed light on this issue.
This is essentially an argument for the counter-factual. Yet it is nearly impossible to find
two or more countries with different budget management systems (one with a cash
budget and others without) but the same level of efficiency in budget management to be
able to distinguish the impact on budget outcome arising from the system versus that
which results from poor management. The difficulty is compounded by the unlikelihood
of being able to control such a comprehensive notion as “efficiency of budget
management” in cross-country analysis.
Second, usually the management and implementation aspects of a budget system
are inextricably linked, so that the question of separating the two is generally an
academic exercise. Indeed, in the case of Zambia, there are inherent features in cash
budgeting that directly contribute to a weakening of budget discipline and budget
implementation, so that disentangling them becomes very difficult. These characteristics
include the following: (1) the fact that the cash bud get system replaces rules-based
procedures with a discretion-based system. Although the budget would normally provide
a framework or set of parameter in which normal cash management occurs, cash
budgeting in Zambia gives decision-makers wide latitude to respond to “emergency”
needs and to authorize supplemental spending at will. Instead of merely managing the
timing of releases, cash budgeting has permitted MFNP to change the overall allocation;
(2) transparency in cash release decisions is generally missing. Because decisions are
made by a small committee, without clear criteria it has been impossible for line
23
ministries to know financing may have been cut, and virtually impossible to anticipate
future funding levels. This lack of transparency in the decisionmaking process also opens
it up to lobbying by individual ministries; and (3) in Zambia the delay in reporting on
expenditures compounds the lack of transparency. Furthermore, it reduces the ability to
incorporate changes into the next year’s budget.
Third, in its simplest sense, a cash budget system implies that the government (or
its agencies) cannot spend more than its (their) revenue. A perfect implementation of this
system should then result, in the aggregate, in a zero budget deficit for the government.
In a perfect cash budgeting world, the problems encountered by the budget system should
be limited to those of revenue projection and the difficulty of smoothing-out revenues
across periods. Its benefits, on the other hand, should be in eliminating large deficits. In
the case of Zambia, while the introduction of the cash budget system has carried with it
the known risk of poor revenue projection, which has led to rigidities in cash releases, it
has not been able to eliminate or even significantly reduce the public sector deficit. The
only accomplishment of cash budget system thus far has been to introduce a false sense
of security, thereby to serve as an instrument to postpone hard fiscal choices and reforms.
Needless to say, this is yet another example of how a budget system may not only have
implications for management, but also for policy choices.
It must be acknowledged that poor budget formulation is also a problem in
Zambia and contributes to some of the variances observed between budget and actual
expenditure. However, this paper takes as an underlying assumption that the budget
approved by Parliament – whether well designed or not – fundamentally represents the
public spending priorities of the nation and should be implemented as such. It follows
that budget execution procedures should be designed to encourage faithful execution of
the budget law, and not undermine it. Yet, the lottery that essentially results from
implementation of the cash budget, undermines the credibility of the Yellow Book that
would normally come from enforcement of hard budget constraints, and it perpetuates an
ineffectual prioritization during budget preparation. In short, the cash budget has
facilitated poor budget planning by enabling important decisions to be deferred until
budget execution, and beyond the immediate view of Parliament.
Even if some component of social and/or economic sector spending is relatively
unproductive, its inclusion in the budget implies that Parliament intended for the program
or activity to be carried out – resources levels not being an issue. It is in budget
preparation that those programs should be reviewed and national priorities debated, and
that by Cabinet and Parliament, but not during budget execution by a small committee of
MFNP. Obviously, revenue shortfalls or national emergencies can lead to some shifting
in spending priorities during the year. In those circumstances, it is wholly appropriate for
Government and Parliament to amend the budget so that the highest priority activities are
protected. This is not what occurs with cash budgeting in Zambia.
.
24
VI.
CROSS-COUNTRY COMPARISONS OF THE IMPACT OF CASH
BUDGETING
The negative impact of cash budgeting on poverty reduction is not unique to
Zambia. At one time or another, cash budget systems have been used by countries as
diverse as Bosnia-Herzegovina, Guinea, Peru, the Russian Federation, Tanzania, and
Uganda.
Uganda and Tanzania have budgeting traditions similar to Zambia’s (due to the
British influence), and both have used a cash release system for years. Their experiences
with cash budgeting are instructive and help demonstrate that Zambia is not an isolated
case.
The analysis and assessment of the experience in these two countries shows some
amazingly close similarities with the situation in Zambia in five key respects
•
Introduction of cash budgeting coincided with achievement of some macroeconomic stabilization. A major factor was improved expenditure controls as a
result of better information and changes in payments systems;
•
Implementation of the cash budget had the same damaging side-effects on the
efficient use and allocation of government resources; i.e creating large,
unpredictable monthly fluctuations in expenditures and a shift in expenditures
from socially and economically important ministries to relatively un-productive
activities;
•
The large monthly fluctuations in cash releases, particularly for RDC, stimulated
over-commitments and arrears;
•
The extent to which damages caused by the cash budget can be minimized or
maximized depends on personal powers of officials in charge of the budget.
Because these officials can change, for example, when there is a cabinet shuffle,
the effects of the cash budget can also vary from year to year; and
•
For different reasons, once the cash budget is adopted, it is likely to stay.
Given these similarities, it is not surprising that many statements in reports
dealing with the budget situation in these two countries apply equally well to Zambia.
One example may suffice. Overall, in each year, a large share of the original budget did
not get implemented – many ministries facing arbitrary cuts – while considerable sums
were spent despite not having been part of the original budget. The bulk of the increases
between estimates and revised budget are geared towards relatively ‘un-productive’
expenditures: allowances, travel, transfers, board and legal expenses. The costs of
additional unbudgeted expenditures are large. They disrupt the budgeted expenditure
25
programme, undermine the institution of the budget as a means for allocating public
resources and add volatility to aggregate expenditure through the year”10
The similarities abound, even though the systems in place in the three countries
are not exactly the same. The Ugandan system, introduced in 1992, is the most flexible. It
is not rule-based, i.e. there is no strict, formal link between monthly revenues and
expenditures as in Zambia, and it allows borrowing from the Central Bank within prudent
limits. The joint Central Bank/Ministry of Finance Cash Flow Committee meets monthly
to discuss and assess the overall financial situation, based on the “cashflow” a set of
monthly fiscal tabulations of resources and expenditures taking into account recent and
expected revenues. The Central Bank plays an important role in this Committee.
However, its recommendations are not binding and the final decision on the next cash
release is that of the Ministry of Finance’s (MOF) alone. As could be expected, this
increased flexibility resulted in somewha t less monthly fluctuations as experienced in
Zambia, but the difference is not striking. While more flexibility also provides more
possibilities for abuse, the MOF pursued decidedly restrictive expenditure policies and
Central Bank borrowing remained limited. In this endeavor it received strong support
from the highest authorities, with the President consistently backing up MOF in disputes
with line ministries about additional spending requests.
An important means of controlling expenditures in Uganda is the regulation that
individual line ministries can not print their own checks. This is done exclusively by
Uganda Computer Services, which is under the administrative control of MOF. Similar to
the rules in Zambia in order for payments to be made, line ministries first have to submit
a list of payments they plan to make, except that in the Ugandan system that list is sent to
MOF rather than the Central Bank. MOF then conducts a pre-audit to verify that the
ministry has sufficient funds to cover the expendit ure.
The Tanzanian system, introduced in May 1994, strengthened in early 1996, is
set up more like the Zambian one with a strict link between cash releases and
expenditures. Monthly cash releases are largely determined by the amount of revenues
collected during previous months 11 , thus avoiding the need for any overdraft from the
central bank. The main difference is that monthly cash releases are based on three-month
moving averages of revenues to smooth-out the monthly flow of cash released to line
ministries.
Both systems suffer from large volatilities in monthly cash releases that affect the
efficiency of resource use, i.e. the ability of expenditure programs to be delivered in a
timely and cost-effective manner. Opinions differ to what extent these volatilities were
created by the cash system or reflect more fundamental problems of lacking budget
discipline, predating the cash-budget period (see below). Volatility appears to be
somewhat less pronounced in Uganda than in Zambia. Like in Zambia, in the Tanzania
10
11
“Macroeconomic Management in Uganda” IMF Working Paper by Mark Henstridge, September 1998
as is the case, de facto, also for a large part of Zambia’s cash releases.
26
system different priorities are attached to different expenditure categories for determining
monthly cash releases with the highest accorded to debt service, followed by personal
emoluments and ending with other charges(OC), including recurrent developmental
charges (RDC), transfers and counterpart funds for development projects. As a result,
RDCs and capital expenditures bear the brunt of any month to month squeeze and, thus,
fluctuate the most. Frequent disruptions in cash allocations for OC disrupt the efficiency
of public services in a damaging way. The MOF decides allocations of OC among budget
heads with little transparency and predictability and without consultation with the line
ministries. Some ministries, however, appear to be particularly successful in securing a
high share. Uganda also prioritizes expenditures by category, protecting debt service and
personal emoluments from most monthly fluctuations. In addition critical social
programs, as decided by the President, also enjoy first priority status.
In both systems heavy re-allocations of expenditures during the course of budget
implementation are the rule, leading to skewed and distorted expenditure composition
compared to original budget estimates. Monthly cash releases are decided with little
attention paid to original budget estimates, constantly overriding the budget as set out in
the annual appropriations. Thus, like in Zambia, the annual budgets in Uganda and
Tanzania have largely been replaced by the monthly cash release with the same negative
consequences. As pointed out in one assessment, too many “irresistible” new expenditure
policies (including large salary increases) are approved on an ad hoc basis during the
course of the year in ways that seem neither transparent nor consistent with
Government’s expressed policy priorities; many for relatively un-productive activities.
To compensate, line ministries for which adequate funding is a priority for future growth
– such as education and agriculture - have not been authorized to spend the full amount of
their original estimates. As a consequence, long-term delivery of public services in the
social and economic sectors is affected, levels and standards of public services suffer, and
productive efficiency is damaged. Further distortions are introduced by the accumulation
of arrears, caused by unauthorized expenditures often in contradiction with long-term
government priorities. Again, opinions differ about the extent this phenomenon is the
result of the cash budget, has only been facilitated by the cash budget, or is due entirely
to pre-existing factors, such as the easy way supplementary budgets are introduced with
little concern for set long-term priorities.
In both systems, over-commitments and arrears are serious problems. Arrears are
used as a de facto financing mechanism when monthly cash releases prove insufficient.
Their emergence are reflections of cash rationing and cash release fluctuations under the
present systems. To reduce over-commitments, a regulation was introduced in Uganda in
1997 stipulating that no commitment was valid unless approved by MOF who has to
ensure that it can be paid for. In Tanzania, a measure was proposed in 1998 under which
prior recording of a commitment with MOF would be a precondition to entertaining a
payment claim. To be honored by the Government, all local purchasing orders (LPOs)
above a certain threshold would need to be countersigned by an Account Examiner
outside the treasury system; these measures to be widely publicized to forewarn private
suppliers. The possibility to repay arrears within the same year by reducing the culprits’
monthly cash releases was rejected as it would only exacerbate the budget problem and
old arrears would simply be liquidated at the cost of accumulating new ones. However, it
was recognized that there is no satisfactory answer to the problem. Within the context of
27
a cash budget system, where the level of total expenditures is given, any repayment of
arrears in a given month or year will reduce the amount of resources available for regular
cash releases and, hence, potentially will stimulate the creation of new arrears 12 There
simply is no way of repaying arrears without taking money from someone or hurting
someone, irrespective of whether the repayment is made in the year the arrears were
created or at any later time.
As in Zambia, the introduction of cash budgeting in Uganda and Tanzania
coincided with achievement of some macroeconomic stabilization. Although there are
minor differences in the way the systems were implemented, their impact on poverty
reduction activities was similar. The magnitude of the impact, of course, depends on the
quality of the decisionmakers who make the ad hoc decisions, as a common feature of the
cash budget is that is gives wide discretion to those who administer it. As noted above,
fiscal control benefits—which had been the justification for cash budgeting—were
undermined by the fact that arrears were used as a de facto financing mechanism when
monthly cash releases proved insufficient.
VII.
REDUCING THE HARMFUL EFFECTS OF CASH BUDGETING
Although the macroeconomic crisis that spurred the introduction of the cash
budgeting system in Zambia in 1993 has long since subsided, the system remains in
place. Fiscal control and accumulation of arrears persist, but Zambia has clearly achieved
some stability relative to the period before the cash budget. To some, the continued
fragility of the country’s economic and financial situation is a prime reason not to replace
the cash budget altogether. In addition, they argue, debt burdens remain high and the
capacity to design and implement a strategic framework for budget execution is weak.
Although these arguments are valid, as the negative effects of cash budgeting become
more obvious, it has become increasingly doubtful whether the macroeconomic
advantages continue to be worth their price.
By its very nature the cash budget helped encourage a weakening in budget
discipline. Decisions debated and approved by Parliament were reopened for
consideration by a small committee within the Ministry of Finance and National
Planning. As a result, expenditure decisions could be driven by immediate, short-term
needs rather than by long-term, strategic considerations. The absence of clear guidelines
on how monthly cash releases were to be determined made it difficult for committee
members to resist such requests.
Restoring the budget’s focus on public service delivery to the poor ultimately
requires reestablishing budget discipline. This is a long-term process that starts with
introducing a transparent and rule-based cash allocation system based on the budget and
reflecting its long-term priorities. Four measures could help achieve this change:
12
This likely was the case in Zambia in 1998 and 1999 when regular budget expenditures had to be
reduced by 9-10 percent because of the need to repay arrears accumulated during earlier years,
increasing the temptation to create new arrears (para. 4.47)
28
•
To improve efficiency in the use of government resources, the time period covered by
each cash release could be gradually extended from one to three and eventually six
months, allowing line ministries to expand their planning horizons. The cash release
operation could be split into two separate steps: publication of a cash release plan and
execution of the actual release each month according to the plan.
•
To improve the allocation of resources, cash releases could be linked much more
closely to the budget, replacing the present ad hoc system by a rule-based system that
maintains long-term government priorities even when overall expenditures have to be
reduced. If the budget itself needs to change to adopt to new circumstances, it should
be done in a fully transparent and formal way.
•
To make overcommitments less attractive, the Ministry of Finance and National
Planning should strictly enforce its mandate to cut back new cash releases to
overcommitted budget heads and use these funds to repay the corresponding arrears.
Doing so would also reduce any expectation ministries may have about obtaining
additional resources through the cash budget.
•
To improve the transparency of the system, the Ministry of Finance and National
Planning could publish more detailed statistics on quarterly expenditures, linking
actual expenditures by ministry (budget head) and major expenditure category to
original budget estimates. Doing so would reinforce the commitment to execute the
budget as planned.
The key to an improvement of the existing budget management system in Zambia
is realistic budgeting and fiscal discipline to execute the budget. In the longer term, the
current cash rationing system has to be replaced by a more conventional budget system,
based on a strategic policy and expenditure framework in which line ministries are
provided with greater responsibility for decisions about the allocation and use of
resources. An important objective of this framework is to increase the predictability of
both policy and funding, so that ministries can plan ahead and programs can be sustained,
while providing line agencies with a hard budget constraint and increased autonomy.
Moving to such a system would increase incentives for the efficient and effective use of
funds.
There is general agreement that this new system should take the form of a
Medium-Term Expenditure Framework. Such a system links government priorities and
the budget within a sustainable spending envelope, highlighting the tradeoffs between
competing objectives and increasing transparency and predictability during budget
implementation. At the line ministry/agency level, this system would link strategy,
activities, budget, and progress indicators in a meaningful way. While it is difficult to
provide a realistic timetable for introducing the new budget system, through the Bank’s
support under an existing adjustment credit, Zambia has begun to take concrete steps to
do so.
29
VIII. CONCLUSIONS
As shown here, there can be very real conflicts between the macroeconomic
stabilization strategies pursued by a country and the overarching goals of improving
service delivery to the poor and reducing poverty. To restore fiscal discipline quickly,
Zambia and other countries have adopted a cash rationing system. While the form the
system takes may vary from country to country, all cash budgeting systems have inherent
features that weaken budget discipline and open up the normal budget process to ad hoc
decision making that has nothing to do with the long-term strategic priorities of the
government. Moreover, even after some degree of stabilization has taken place, the cash
rationing system continues to remains, along with the deleterious practices it introduced.
In Zambia the record clearly and unequivocally shows the deleterious impact of
cash rationing on ministries that promote economic growth and social development. As
the fiscal statistics show, social services and economic services ministries found their
cash releases substantially reduced relative to the amounts in the budget, and cash
releases were subject to very significant monthly volatility.
The impact on service delivery is apparent. Under- funding exacerbates an already
strong bias toward personnel inputs at the expense of operating goods and services. In
addition, work plans that may have been predicated upon a certain level of funding and
service delivery are suddenly destabilized. The lack of predictability as to what the new
funding level will be makes it extremely difficult for managers to plan appropriate
modifications to their service delivery model.
The greatest problem facing line ministries and budge t heads is the unpredictable
monthly volatility of cash releases. Where cash rationing decisions are made on a
monthly basis, line ministries have too short an horizon over which to plan. The shortage
of key inputs can force ministries to engage in unproductive activity or to obtain
resources on credit, thereby accumulating arrears.
Not only does cash budgeting fail to encourage a focus on poverty reduction and
service delivery, it fails to achieve fiscal discipline as well. While actual cash payments
are more tightly controlled under the cash budget system, the underlying commitments of
the government can easily escape scrutiny. Zambia, like Uganda and Tanzania, has
shown that arrears build- up remains a challenge for the government well after the
imposition of cash budgeting.
Thus, while the original noble intentions of a cash rationing system is to bring
spending in line with resources, in practice, the system fails to achieve both the desired
level and composition of public spending because it cannot replace fiscal discipline and
because it brings ad-hoc, discretionary policy decisions into the public expenditure
management arena. The case of Zambia shows that countries must carefully consider the
tools they choose to achieve macroeconomic stabilization, assessing their likely impact
on service delivery to the poor. Cash budgeting strategies present a potentially high risk
to poverty reduction when the practice encourages decision making that takes place
outside of a clear framework of rules that carefully link decisions back to the budget and
to long-term development priorities, when decisions are made on an ad hoc basis each
30
month without an adequate planning horizon for line ministries to respond and adjust,
when there are few sanctions on powerful ministries for using overcommitments as
leverage to gain funding, and where there is little transparency in the resulting cash
release decisions.
31
Reference
Bolnick, Bruce.1995. “Establishing Fiscal Discipline: The Cash Budget in Zambia”,
mimeo.
Henstridge, Mark. 1998. “Macroeconomic Management in Uganda”, IMF Working
Paper.
Dinh, Hinh, Heinz Bachmann, Abebe Adugna and Mushiba Nyamazana. 2000. “Cash
Budget in Zambia: Stabilization versus Growth and Poverty Reduction”, World Bank
AFTP1..
Kitabir, Damoni. 1996. “The Budgetary Process in Uganda: the Legal Framework,
Institutional Arrangements, and Implementation” Paper prepared for the AERC Budget
Process Project.
Republic of Zambia. “Estimates of Revenue and Expenditure” (“Yellow Book”), 1995,
1996, 1997, 1998.
Republic of Zambia. “Financial Reports” (“Blue Book”), 1995, 1996, 1997, 1998.
Stasavage, David and Dambisa Mayo. 1999. “A Cash Budget: A Cure for Excess Fiscal
Deficits (and at what cost)? Mimeo.
World Bank. 1992. “Zambia Public Expenditure Review”, AFTP1.
World Bank. 1997. Tanzania Public Expenditure Review. Vol.1, Ch. 6.
World Bank. 1998. “Zambia Fiscal Management Report”, No. 18552 ZA, AFTP1.
World Bank. 1998. “Zambia: Money, Bond, and Foreign Exchange Markets”, AFTP1.
World Bank. 2001. “Zambia Public Expenditure Review”, AFTP1.
.
32
ANNEX TABLE 1
Zambia: Selected Macroeconomic Indicators
1992 1993 1994 1995 1996 1997 1998 1999 2000
Real GDP Growth, %
-1.7
6.8
-8.6
-2.5
6.6
3.3
-1.9
2.0
3.5
Inflation (end-of-period), %
191.1
143.3
34.2
46.0
35.2
18.6
30.6
20.6
30.1
Inflation (period average), %
191.4
187.1
53.3
34.9
43.1
24.4
24.5
26.8
26.1
Domestic Budget Balance,
cash, % of GDP
(4.0)
(4.2)
(1.3)
0.3
1.3
1.1
0.4
0.4
(3.7)
Overall Budget
Surplus/(Deficit), cash, % of
GDP
(1.7)
(5.6)
(6.8)
(3.8)
(5.4)
(4.1)
(8.0)
(7.7)
(7.8)
GRANTS, % of GDP
10.2
8.0
5.3
5.2
6.1
5.1
6.6
8.1
6.4
(11.9)
(13.6)
(12.2)
(9.0)
(11.5)
(9.2)
(14.6)
(15.7)
(14.2)
Overall Budget
Surplus/(Deficit), before
grants, cash, % of GDP
Source: World Bank Data Base
33
34